小狗hammer

我遇到的问题： 我将google mediapipe hand的模型移植过来，图像处理部分用MNN提供CV库进行预处理 google mediapipe hand一共有两个模型，一个是[palm检测](https://github.com/google/mediapipe/blob/master/mediapipe/modules/palm_detection/palm_detection.tflite) 一个是[hand检测](https://github.com/google/mediapipe/blob/master/mediapipe/modules/hand_landmark/hand_landmark.tflite) 问题出在第二个模型的预处理上 **安卓报的错误是：** `2021-09-08 18:17:30.954 26518-26518/? A/DEBUG: signal 11 (SIGSEGV), code 2 (SEGV_ACCERR), fault addr 0x7c6a0ca06d 2021-09-08 18:17:30.954 26518-26518/? A/DEBUG: x0 b400007c6a059040 x1...

const uint8_t* data 也就是640x480 RGB数据，我通过MNN::CV转换成224x224 另外我发现报错的手机是android-11,我另外有个手机是android-10，这个是10是可以运行的

Thank you. I found that the such a **big difference** is caused by DSP. I don't think it has anything to do with AIMET. There seems to be a bug...

Update ResNet18 example: ``` class resnet18_removefc(nn.Module): def __init__(self): super().__init__() self.model = resnet18(pretrained=True) def forward(self, x): x = self.model.conv1(x) x = self.model.bn1(x) x = self.model.relu(x) x = self.model.maxpool(x) x = self.model.layer1(x)...

Sorry, it's been a long time. I don't remember the exact results at that time. The conclusion is that 1. there is a problem with SNPE linear layer. 2. SNPE...